• Welcome to the Thea Render Community
    Feel free to create an account and start interacting with the community. Please make sure that you check your junk/spam folder for the confirmation email.

Thea for Rhino + Linux TheaNode feedback

nart

New member
Hello TheaNode devs-

I've been evaluating TheaNode with the Rhino eval license.

It is excellent to hear that the nodes are unlimited no charge, and you plan to support the big three OS's. Hopefully it will stay that way.

I'm having some issues though, I think related to:

Stopping a render takes a very long time...

I guess this is a function of network speed and resolution, as I can see the 'scanner' is saturating its receive channel during the 'stopping' phase. I guess all the nodes are sending back their final images, but it is also breaking some nodes, perhaps because of some kind of timeout? I don't mind that it takes a few minutes to stop, but I also don't think it should cause a node to exit. After a stop, I've seen the nodes go back to one of three states:

1 - Node goes back to 'REQUESTING PROJECT FROM SERVER: ...
2 - Final line is written like 'Device #0: 1029x4 s/p', but never prints the 'finished' line or goes back to requesting
3 - Process exits

State 1 will usually pick up the next project.
State 2 sometimes picks it up. Also, the node may be using CPU at this time, after Darkroom has completed.
State 3 requires a manual restart of the process.

Is there a parameter to turn on verbose or debug logs so I can see what is happening?

Thanks,

Nathan
 
thanks for trying TheaNode and for the feedback. Sorry for the delay, I will answer you when I am available the next month, at the start on June.
 
I had some time over the weekend so I did an experiment on this. Reporting results here.

I pulled 3 nodes from the rack and set them up local to my workstation, connected via static gigabit ethernet. This bypasses my wireless channels (to the rendernode location), firewall / VLAN traversal, and any DHCP/DNS issues. Note that my network works for everything else, so I don't think there is an issue with it, but as the first troubleshooting step I thought it best to rule it out first.

The nodes are linux containers (debian) hosted via a proxmox cluster. The virtualization is the same as before, although I couldn't use the same physical hosts. I spun up 12 rendernodes, 4 per physical host, and gave them 8 cores each for 100% host cpu utilization. All networking was done with manual IP assignments.

At full gigabit speed, I saw one node fail to stop during all testing.

I throttled each container to 300mb/s and started to see issues with stopping nodes. Starting/Stopping took longer, as expected.

Even more issues with 54mb/s, pretty much unusable. Not all nodes starting, and stopping causing hangs/exits.

So whatever the problem is, it seems to be tied to bandwidth.
 
Back
Top